Abstract
The SARS-CoV-2 outbreak from October 2020 through February 2021 was the largest outbreak as of February 2021, and timely information on current representative prevalence, vaccination, and loss of prior antibody protection was unknown. In February 2021, the South Carolina Department of Health and Environmental Control conducted a random sampling point prevalence investigation consisting of viral and antibody testing and an associated health survey, after selecting participants aged ≥5 years using a population proportionate to size of South Carolina residents. A total of 1917 residents completed a viral test, 1803 completed an antibody test, and 1463 completed ≥1 test and a matched health survey. We found an incidence of 2.16 per 100 residents and seroprevalence of 16.4% among South Carolina residents aged ≥5 years. Undetectable immunoglobulin G and immunoglobulin M antibodies were noted in 28% of people with a previous positive test result, highlighting the need for targeted education among people who may be susceptible to reinfection. We also found a low rate of vaccine hesitancy in the state (13%). The results of this randomly selected surveillance and associated health survey have important implications for prospective COVID-19 public health response efforts. Most notably, this article provides a feasible framework for prompt rollout of a statewide evidence-based surveillance initiative.
Despite the unprecedented scientific effort to alleviate the SARS-CoV-2 pandemic, many epidemiologic questions remain unanswered. Further studies are necessary to clarify antibody decline, the disease burden of pediatric infection, and disparities of infection between racial and ethnic groups.1-3 Interagency collaborative efforts to produce isochronous reporting of population-level prevalence will remain critical as the pandemic continues. A statewide study in Indiana showed the value of population point prevalence–based information conducted in April 2020. 4 The study provided context for the importance of statewide randomized sample population studies that help obtain generalizable information rather than information on people seeking care for known exposures. This type of study provides information about new potential transmission methods and helps improve public health studies to approach the pandemic at a state level.
The United States reached 30.4 million cumulative cases of SARS-CoV-2 on March 31, 2021, and was the country with the greatest number of COVID-19 cases and deaths after the winter peak. 5 On December 14, 2020, the United States initiated the first rollout of COVID-19 vaccines to the population with no expectations on what the uptake would be; 54 607 041 (16.6%) people were fully vaccinated as of March 31, 2021. Winter peaks are of concern because this seasonal pattern often yields new emerging variants. As such, vaccine uptake is important to reach herd immunity. As of March 31, 2021, South Carolina had 447 507 cumulative cases of SARS-CoV-2. 6 The SC STRONG project is a collaborative effort led by the South Carolina Department of Health and Environmental Control (SC DHEC) with multiple community partners to provide quarterly cycles of free COVID-19 viral and antibody testing for selected participants in South Carolina. The objective of this study was to describe the results of testing in our second cohort in February 2021.
Methods
We generated sample size estimates using Stat-Calc’s population survey calculator, EpiInfo version 7.2.3.1 (Centers for Disease Control and Prevention [CDC]). We determined the required sample size of 2294 (range, 1398-3649) based on COVID-19 national population infection estimates of 16% (range, 9%-31%), 7 the current South Carolina state population, and a 1.5% acceptable margin of error. To achieve a sample size ranging from 1398 to 3649 residents, we increased selection to accommodate anticipated response rates for a final total of 117 563 South Carolina residents. These 117 563 residents were randomly selected using the population-proportionate-to-size cluster sampling method, comprising 30 residents each from 3922 geographic clusters (zip codes). Adult residents were selected from a comprehensive third-party direct-mail marketing listserv (MailersHaven). All selected participants, and 1 household member aged ≥5 years of the initially selected participants’ choosing, received a mailed invitation letter. We used joint partner press releases and dedicated project resources, including Spanish translation, hotline, email, and website (https://scstrong.sc.edu), to encourage participation. We sent weekly reminders using an integrated pre-scripted chatbot (Conversa Health) to solicit follow-up participation.
Participants were asked to complete an electronic survey and provide biological samples for SARS-CoV-2 virus and immunoglobulin M (IgM)/immunoglobulin G (IgG) antibodies at 70 partnering sites across the state. Laboratory assays were performed at the SC DHEC Public Health Laboratory, and point-of-care antibody tests were performed on-site with results reported to the laboratory. The SC DHEC and University of South Carolina institutional review boards determined surveillance protocols to be public health surveillance, human subjects research exempt. A diverse, representative consortium at the University of South Carolina’s Patient Engagement Studio (https://hsc.ghs.org/research/pes) reviewed all surveillance protocols and outreach documents.
We computed poststratification weights based on public health region, race and ethnicity, age, and sex, followed by weighted logistic regression to compute estimated incidence proportions for active infection and seroprevalence. We used unweighted univariate logistic regression to compute raw associations between survey responses and disease status. We considered P < .05 to be significant.
Results
From February 1 through March 4, 2021, a total of 1917 residents completed a viral test, 1803 completed an antibody test, and 1463 completed ≥1 test and a matched health survey. The response rate was 1.2% (1463 of 117 563 randomly selected residents) for the survey on personal risk factors and vaccine-related questions. The estimated state incidence was 2.16 (95% CI, 1.47-3.18) per 100 people and seroprevalence was 16.4% (95% CI, 14.4%-18.5%) (Table 1). Children and adolescents had the highest estimated incidence: 5.71 (95% CI, 1.88-16.11) per 100 residents aged 5-19 years. The highest seroprevalence was among Hispanic residents (32.9%; 95% CI, 19.5%-49.8%), non-Hispanic Black residents (20.6%; 95% CI, 14.4%-28.5%), people aged ≥60 years (21.6%; 95% CI, 19.3%-25.4%), and people living in the Upstate public health region (18.5%; 95% CI, 15.1%-22.3%). A median viral polymerase chain reaction (PCR) detection duration of 29 days (range, 8-56 days) was noted among people with a current and previous PCR test result (n = 14). Antibody loss occurred in 28.0% of residents with a previous positive test result, with a median of 136 days (range, 19-326 days) between current negative antibody test result and previous PCR or antibody-positive test result (42 of 162).
SARS-CoV-2 point incidence and seroprevalence, by region and demographic characteristics, South Carolina, February 2021
Viral testing was performed using TaqPath COVID-19 RT-PCR Multiplex Assay (Thermo Fisher Scientific).
Poststratified for nonresponse, and incidence rate is per 100 residents aged ≥5 years.
Antibody positive was defined as being immunoglobulin G (IgG) and/or immunoglobulin M (IgM) positive on the SARS-CoV-2 IgG/IgM assay (Abbott Industries), COVID-19 IgG/IgM Rapid Test Cassette (Healgen Scientific LLC), or Assure COVID-19 IgG/IgM Rapid Test Device (Assure Tech).
Poststratified for nonresponse and adjusted for specificity (0.996) and sensitivity (0.975).
Zero count of polymerase chain reaction–positive test results among participating Hispanic people, calculated an estimate incidence proportion of 0.000000007 (range, 0-1).
People working in a nursing home or as a frontline medical care worker had 7.0 and 4.6 times higher odds of receiving a positive test result for virus or antibody, respectively (Table 2). Being aged ≥70 years was associated with 2.2 times higher odds of receiving a positive test result. Lastly, having a close family member or friend with COVID-19 or currently experiencing symptoms was associated with 2.8 or 1.7 times higher odds of receiving a positive test result, respectively. Having COVID-19 symptoms was significantly associated with having either a PCR or antibody-positive test result; however, 20 of 33 (60.6%) people with a PCR positive test result were asymptomatic, suggesting a potential symptom latency.
COVID-19 risk factors and vaccine-related questions among all survey respondents (N = 1463) and stratified by COVID-19 status, South Carolina, February 2021
Total number of affirmative responses divided by the total number of survey responses for that question.
Completed viral RNA, immunoglobulin M, and/or immunoglobulin G antibody surveillance-related diagnostic assay.
Significant at P < .01.
Finding is not significant (ie, P > .05).
n = 1396.
n = 1146.
n = 1456.
n = 925.
Continuous variables were analyzed by linear regression, which does not compute odds ratios.
n = 188.
n = 1455.
n = 987.
Denotes reference category in logistic regression.
Response shown is “agree” compared with “neutral” or “disagree.”
n = 1452.
n = 1450.
n = 1447.
We found no significant differences between vaccine perceptions or attitudes by testing status. However, overall, the rate of vaccine hesitancy was low (13.0%), and most people reported planning to receive the vaccine as soon as they were eligible (87.0%) (Table 2). Vaccine uptake was high, with 19.0% self-reporting receipt of their first vaccine and 4% reporting receipt of their second vaccine at the time of sample testing and survey completion. Among survey respondents, most felt COVID-19 vaccines were safe (71.0%) and effective (69.0%). In addition, most respondents felt confident in the design and development of the vaccines (73.0%) and the regulatory approval process of the vaccines (71.0%).
Discussion
To our knowledge, this article is one of the few large US-based statewide representative sampling to have been performed during the SARS-CoV-2 pandemic. This sampling identified an estimated incidence of 2.16 per 100 residents and seroprevalence of 16.4% for SARS-CoV-2 in South Carolina. This estimated case burden is 2 times higher than the reported number of diagnosed cases by SC DHEC, 6 highlighting the possible role of asymptomatic transmission and failure to seek testing during active infection. Consistent with previous epidemiologic studies, Hispanic and Black populations had the highest case burdens. 2 Although Hispanic (32.9%) and non-Hispanic Black (20.6%) residents had the highest SARS-CoV-2 seroprevalence, these findings were not significant by logistic regression, suggesting infection earlier in the pandemic likely occurred in these demographic groups. Report of COVID-19–related symptoms was significantly associated with having either a positive PCR and/or antibody test result, suggesting prolonged symptomatic disease even among people who had an antibody-positive test result but a PCR-negative test result. Interestingly, we found a higher PCR incidence in our pediatric population (defined in our study as aged 5-19 years) but a higher seroprevalence in our older population (defined in our study as ≥60 years). This finding might reflect a recent return to in-person school mandate 8 or the general nature of high infection burdens among pediatric populations. 2
Antibody loss was noted in 28% of our population that had a previously positive test result, which is notably higher than the 6% noted by CDC. 9 Our higher seronegative percentage is likely due to the greater duration between our 2 studies: a median of 136 days since previous positive test result (current) versus a mean of 63 days for CDC. 9 Conversely, our survey identified 19% of the population already had been vaccinated or were in the process of getting vaccinated for COVID-19, consistent with SC DHEC’s vaccination reporting. Vaccine hesitancy was low among our surveyed population, in contrast to current polls and literature. 10 This promising finding suggests that most residents elect to receive the vaccine when eligible and that confidence in the vaccines’ efficacy, safety, development, and regulatory processes exists. Our representative survey’s finding of low COVID-19 vaccine hesitancy is consistent with a separate unpublished SC DHEC vaccine survey, which occurred at the same time as the current survey, which found that 80.5% of residents planned to receive the vaccine when eligible.
The impact of vaccination on antibody positivity (n = 324 total antibody positive) in our surveillance cohort is unknown. At the time of testing, 4% (n = 64) of residents reported having their second vaccination dose, with a median duration between last vaccine and sample testing of 14 days (range, 0-49 days). Abbott’s SARS-CoV-2 IgG assay detects immunoglobulin class G antibodies to the nucleocapsid protein of SARS-CoV-2, not a target of available COVID-19 mRNA vaccines (personal communication, Joe Harvey, Abbott, March 17, 2021). In contrast, Abbott’s SARS-CoV-2 IgM assay detects immunoglobulin class M antibodies to the spike protein of SARS-CoV-2 targeted by available COVID-19 mRNA vaccines (personal communication, Joe Harvey, Abbott, March 17, 2021). Healgen and Assure assay manufacturers did not respond to our information request. It should be noted that previous vaccination might have affected the findings of increased odds of receiving a positive SARS-CoV-2 test result among nursing home and frontline health care workers and people aged ≥70 years. However, 27 of 64 (42.2%) people with a second dose of the vaccine were IgG antibody positive, suggesting naturally acquired immunity in this group. We anticipate the high positivity rates among these 2 occupational groups is likely a mix of previous vaccination and occupational exposure, which has been upwards of 22.5% in previous studies. 11
Limitations
Our randomly selected surveillance approach had several limitations. First, we had a low response rate, which could be the result of inaccurate or outdated resident addresses; the politicized nature of the pandemic, which may have hindered motivation to participate in a state health department activity; access to participating sample collection sites; or other causes. Although we experienced a low response rate, this rate is consistent with other public health reports. 4 Second, given the cross-sectional design, this study of residents of South Carolina is not generalizable to other states or other pandemic time points. Third, a period effect could be another limitation. Because of the unstable political situation in the United States at the beginning of the study, an increased influence based on political opinion could have caused a selection bias that affected participation later in the study. Although we solicitated participation from a randomly selected population, those who sought out participation might have been a highly motivated self-selected population, leading to less randomization in the final analysis. These limitations might explain the lack of age and racial and ethnic representativeness compared with the general South Carolina population (Table 3). We anticipated that we might have challenges, and actions were taken to foster support among young people and those in racial and ethnic minority groups, including the option for finger-prick assays versus full venipuncture for children and pop-up events in neighborhoods with large racial and ethnic minority populations. To statistically accommodate this limitation, we attempted to correct for the nonrepresentativeness of the sample by poststratifying our analysis so that distribution of public health region, race and ethnicity, age, and sex of the poststratified sample resembled that of the state as a whole. 12 However, poststratification is not a perfect solution to the problems presented by a sample that is not fully representative. Vaccine acceptability among the respondent group should be interpreted in context of this potential selection bias, because the current vaccination rate in South Carolina does not reflect the willingness reflected in the survey responses. Lastly, our investigation is not representative of children aged <5 years, institutionalized people, patients in hospitals or rehabilitation centers, or populations from rural environments that did not have access to as many testing facilities as populations in urban areas.
A regression comparative analysis of demographic characteristics of South Carolina STRONG participants and the overall South Carolina population to assess representativeness of the randomly selected population who decided to participate in a survey on COVID-19 vaccine acceptability, February 2021
P < .05 was considered significant.
Conclusion
In South Carolina, randomly selected resident surveillance estimated that only half of COVID-19 infections were officially diagnosed by the state health department before March 2021. An estimated incidence of 2.16 per 100 residents and seroprevalence of 16.4% were identified after the latest pandemic surge. Antibody loss was common and highlights the need for targeted messaging on the value of vaccinations among previously infected people. Although antibody levels do not represent the only immunity against SARS-CoV-2, specific B-cell and T-cell responses have been seen leading to a decrease in severity of disease in subsequent infections.13,14 Notable health disparities exist, with up to 33% of Hispanic and 21% of Black residents seropositive compared with 15% of non-Hispanic White residents. Our surveyed population demonstrated a low rate of vaccine hesitancy, which holds promise to a future achievement of herd immunity through a high vaccination rate in South Carolina and surrounding states. Overall, this surveillance initiative highlights the value of evidence-based public health surveillance during the initial phases of a novel pandemic. Future steps should include the continuation and adaptation of such randomly selected resident cohorts for prospective enhanced state-level public health surveillance.
Footnotes
Acknowledgements
The authors thank the South Carolina residents and clinical partners who participated in this surveillance initiative; our additional partners from the SC STRONG project team: Kenneth R Deans, Jr (Health Sciences South Carolina [HSSC]), Katrina Fryar (HSSC), Jessica Siebenschuh (HSSC), Graham Adams (South Carolina Office of Rural Health), Sarah Cockrell (South Carolina Primary Health Care Association), Nitin Patel (Prisma Health), Darin Thomas (Prisma Health), Roxanne Ancheta (South Carolina Department of Health and Environmental Control [SC DHEC]), Scott Hagins (SC DHEC), Danielle Johnson (SC DHEC), Connor Ross (SC DHEC), O’Neal Best (University of South Carolina [USC]), Madlyn Wills (USC), Madison Collins (USC), Sammy Khalil (USC), Darin King (USC), Sophie Henry (USC), Josephine A. Morrisey (USC), Michael Yan (USC), Amelia Hornaday (USC), Kateryna Parkhomenko (USC), Callie Shirley (USC), Sadie Hildebrandt (USC), Hanson Cowan (USC), and Abbie Drake (College of Charleston); and Ann Blair-Kennedy, PhD, and the Patient Engagement Studio members for their contributions.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received the following financial support for the research, authorship, and/or publication of this article: Funding for this surveillance initiative was provided by the US Department of Health and Human Services (grant # NU50CK000542). The funders had no role in the design, data collection and analysis, decision to publish, or preparation of the article.
